AbstractObjectivePrevious studies have shown either a lack of effect of IRF5 polymorphisms or an association of the IRF5 gene in only a minor subset of rheumatoid arthritis (RA) patients in whom anti–citrullinated protein antibodies (ACPAs) are absent. The present study was undertaken to investigate the role of genetic variation in IRF5 in susceptibility to RA.MethodsNine IRF5 single‐nucleotide polymorphisms (SNPs) were studied in 1,338 patients with RA and 1,342 control subjects in analyses of exploratory and replication sample collections, with stratification according to sex and by the presence or absence of ACPAs, rheumatoid factor, the shared epitope, the 620W PTPN22 allele, and erosions. A meta‐analysis that included results from previous studies was also carried out.ResultsOur findings together with those from previous studies, in a total of 4,620 RA patients and 3,741 controls, showed a significant association of the rs2004640 IRF5 SNP in RA patients as a whole (odds ratio [OR] 0.88, 95% confidence interval [95% CI] 0.83–0.94; P = 6.5 × 10−5 versus controls). This association was stronger in ACPA− patients, but was also present in ACPA+ patients (from 3 sample collections). Further analysis of our exploratory sample collection showed that only patients in the ACPA+ and SE− group lacked an association with IRF5 SNPs. All of the remaining RA patients (ACPA− or SE+) showed a strong association with IRF5 SNPs, which followed a complex pattern of opposing effects mediated by independent haplotypes. The susceptibility haplotype showed an OR of 1.8 (95% CI 1.4–2.3; P = 1.2 × 10−6 versus controls), whereas the protective haplotype showed an OR of 0.76 (95% CI 0.6–0.98; P = 0.046 versus controls).ConclusionIRF5 polymorphisms seem to influence RA susceptibility in a large subgroup of patients, following a pattern of association very similar to that described in patients with systemic lupus erythematosus.